Amazon can make just about anything—Except a good video game
Mike Frazzini had never made a video game when he helped start Amazon Game Studios. Eight years later, he has released two duds, withdrew both from stores after a torrent of negative reactions and canceled many more. For a company that dominates countless areas of retail, consumer electronics and enterprise computing, the multiple failures in gaming show one realm that may be impervious to Amazon.com Inc.’s distinctive business philosophy. It tried to make games the Amazon way, instead of simply making games people would want to play.
Frazzini is an Amazon lifer who came up in the books section of the website, where he endeared himself to Jeff Bezos as a manager there. Conventional wisdom inside the company is that if you can run one business, you can run any other. Amazon’s deep financial resources certainly help. As head of the games division, Frazzini has acquired established development studios and pushed the company to spend nearly $1 billion for the live video streaming website Twitch. Frazzini recruited some of the top names in the video game industry, including creators of the critically acclaimed franchises EverQuest and Portal, as well as executives from Electronic Arts Inc. and other big publishers.
Then, according to numerous current and former employees of Frazzini’s game studios, he ignored much of their advice. He frequently told staff that every Amazon game needed to be a “billion-dollar franchise” and then understaffed the projects, they say. Instead of using industry-standard development tools, Frazzini insisted Amazon build its own, which might have saved the company money if the software ever worked properly. Executives under Frazzini initially rejected charges that New World, an Amazon game that would ask players to colonize a mythical land and murder inhabitants who bear a striking resemblance to Native Americans, was racist. They relented after Amazon hired a tribal consultant who found that the portrayal was indeed offensive, say two people who worked on the project. The game, previously planned for release last year, is now scheduled for this spring.
This story is based on interviews with more than 30 current and former Amazon employees, most of whom spoke under the condition of anonymity citing fears of litigation or career repercussions. A spokeswoman for Amazon declined to comment or make Frazzini available for an interview.
Amazon is spending nearly $500 million a year operating the video game division, two people familiar with the budget say. That amount doesn’t include Twitch or a new project under different management, which is building a service to stream games to a computer, phone or Amazon Fire TV.
To develop games, Amazon tried to bend a creative and collaborative process to its will, and the results offer lessons to Apple Inc., Facebook Inc. and Google, whose efforts so far have been similarly ineffective. Successful video games are a combination of art, entertainment, technology and very large budgets. Big tech companies only really figured out the last two.
Many of the game developers who joined Amazon found themselves repelled by the corporate culture. The company is driven by data, and employees are expected to write six-page documents to get major decisions approved. In game development, on the other hand, a phrase often uttered around the office is “finding the fun.” It refers to altering and polishing small aspects of a game to figure out what makes the experience enjoyable. The results are measured only in emotion, which is why many developers say it’s critical for the people in charge to have experience making games.
At Amazon’s core is a set of 14 leadership principles. They include “customer obsession” and “frugality.” For a company man like Frazzini, they offer a scale by which every member of the team is measured. “If you don’t come in line with that approach, you’ll struggle at Amazon,” says Jason Child, who spent more than a decade in Amazon’s finance department. Adapting to the corporate culture counts for much more than expertise, he says. “If someone is a guru in video games and they go to Amazon, would they be successful? Probably not.”
Jeff Bezos is more of a book guy, but the directive to go into video games came straight from the top. The Amazon chief executive officer indicated he was willing to spend exorbitant sums of money and offer development teams as long as they needed, say three people who worked directly with Bezos. All that mattered was that they make the most ambitious games possible, ones that would draw gamers into the Amazon Prime ecosystem and showcase the technical capabilities of its cloud division. Allowing 10,000 people to play in a single game session was given to the new team as a lofty target. Two projects under this directive became known around the office as the “Bezos games.”
Bezos views games as yet another way to sell subscriptions to Prime and hook customers on its other offerings, including television shows and movies. As other e-commerce companies catch up with speedy free shipping, Amazon has sought to add more perks to Prime to justify the price of membership. Frazzini helped kick off the games initiative in 2012 and soon became boss. He had risen from the books department to run the section of Amazon’s store that sells video games shipped by mail. At first, the new group planned to make games for the Amazon Appstore and release them on Android phones, including the ill-fated Amazon Fire Phone, as well as the Fire TV. By 2014, the company decided to move into games for PCs and eventually consoles. It was a dramatic change. Mobile games can be made by small teams in just a few months, whereas bigger ones take hundreds of people toiling away for years.
Frazzini set up a new game development operation at Amazon’s headquarters in Seattle. They would later call it Relentless Studios, using one of Bezos’s favorite adjectives for the company. To help run it, Frazzini tapped Louis Castle, who founded the influential game company behind the Command & Conquer series.
Amazon hired celebrated developers like Kim Swift, designer of the puzzle game Portal, and Clint Hocking, director of the shooter Far Cry 2. It also formed two more studios in California. The splashy hires kept coming, including Madden guru Richard Hilleman and online gaming pioneer John Smedley. Today, only Smedley remains. All declined to comment.
At first, new recruits thought they were entering some sort of fantasy land. Many were paid double the market rate of other game makers in the area, on top of lucrative packages of Amazon stock that just kept rising in value. Teams had deadlines, but they proved to be flexible, and overtime requests were infrequent, more than a dozen former employees say.
One aspect of working at Amazon felt similar to traditional game companies. The studios cultivated a “bro culture” in which women often weren’t given the same opportunities as men, former employees say. Four female game developers say their worst experiences of sexism in the industry were at Amazon. They shared stories of being ignored and undermined by male executives and say they were eventually driven out of the company. One former employee says male colleagues completely ignored her comments in meetings. Another says a member of senior leadership impeded her career growth after she disagreed with him and that he created new management positions above her and filled them with men.
Other employees registered less consequential complaints about all the company jargon thrown around the office. Frazzini regularly expounded on the Amazon leadership principles. One of those was “Hire and develop the best,” which was flattering to employees but didn’t tell the full story. “The philosophy Amazon takes is: Hiring expertise is secondary to having leaders who follow the Amazon principles,” says Child, the former employee. “Amazon does hire experts from various industries, but then they expect those folks to adopt the Amazon way.”
The game studios even established their own separate set of principles, although the credos frequently changed and sometimes were in tension with one another, say four people who worked there. Each game world should accommodate as many players as possible, yet also be fun to play solo at the same time. They had to be huge financial successes on a Call of Duty scale but also innovative and unlike anything the world had seen before. To experienced game developers, these rules seemed like a surefire way to not release anything.
Amazon didn’t give employees much financial incentive to release anything, either. Most big game companies pay staff bonuses based in part on the critical and commercial response to their games, but Amazon’s stock plan only rewards employees for time spent at the company. That led some to prioritize job preservation over anything else, say three former employees. They say they watched colleagues avoid arguments and only seek to placate bosses like Frazzini, even when they disagreed. (This was in defiance of the Amazon principle “Have backbone; disagree and commit.”)
Frazzini’s lack of experience in video games showed during project review sessions, a standard industry ritual when the boss plays early prototypes and offers feedback. His comments were of the focus-group variety, recalls a former Amazon developer: “Why is it this color?” and “Seems fun. When will it be ready?” On a different occasion, says another developer, the team cringed as Frazzini struggled to differentiate between hyper-polished conceptual footage and live gameplay, a sign he didn’t understand the technology.
Some meetings got sidetracked when Frazzini, armed with the latest VentureBeat article about whatever game was making the most money that month, demanded they chase a new trend, four developers say. The team wound up designing lesser versions of popular games, a desperate strategy laid out in a recent Wired article. Riot Games Inc.’s League of Legends inspired an Amazon project called Nova that was canned in 2017. Epic Games Inc.’s Fortnite led to another Amazon game, Intensity, canceled in 2019. Activision Blizzard Inc.’s Overwatch begot Amazon’s Crucible, which would suffer a similar fate.
Amazon’s most successful gaming property isn’t actually a game. Twitch draws some 26 million people a day to watch live video of other people playing video games. Frazzini helped secure the acquisition in August 2014 for $970 million, which would later prove to be a bargain. Various attempts by Facebook, Google and Microsoft Corp. to build their own live-streaming networks for games have failed.
Soon after the Twitch deal, Bezos devised a plan to closely pair Amazon Game Studios with the video site, say two people who met with the CEO. Twitch would serve as a free marketing vehicle for Amazon games, and the games, in turn, would offer Twitch viewers features they couldn’t get anywhere else.
The two teams held daylong brainstorming sessions, but the summits didn’t amount to much, according to people who attended. One concept they did pursue together was a series of celebrity events to promote their games and sell subscriptions to Amazon Prime. They held one in New Jersey featuring the Clerks filmmaker Kevin Smith. Attempts to land other stars, including Dwayne “The Rock” Johnson, for a second event in Las Vegas were unsuccessful. They invited Smith back and then shelved the series.
The Twitch chief, Emmett Shear, was reluctant to assign his employees to projects for the game studios, two people who worked with him say. One proposal, pushed by Frazzini in 2017, asked Twitch to help build a digital game store, the former employees say. The idea would allow Amazon to avoid paying a 30% sales commission to Steam, the most popular storefront for PC games. Twitch employees weren’t thrilled because Amazon didn’t have any big games with which to attract shoppers. Frazzini’s employees assigned to collaborate on the project protested because bypassing Steam would shrink their market, four former employees say. Withholding a game from Steam, they argued, is the equivalent of refusing to sell a book on Amazon. When the project was due for more staffing resources, Shear resisted, and it was left to languish inside Amazon Game Studios.
Frazzini was more fixated on another project anyway. Amazon’s designers and programmers needed a game engine, a collection of tools used to build games. For most studios, there are really only two options. Epic’s Unreal Engine is Coke, and Unity Software Inc. is Pepsi. Amazon, in effect, decided to make an RC Cola. In 2014, it licensed technology from the German company Crytek for a homemade engine called Lumberyard. Frazzini then assigned a team of engineers to build the engine and released it to the public in 2016 for free. The tools are intertwined with Amazon Web Services, setting up Lumberyard as a way to draw a new class of software developers to the business. Frazzini, who reports to Web Services Chief Andy Jassy, also mandated that all Amazon games be built with Lumberyard, rather than pay for Unreal or Unity.
Lumberyard became a bogeyman around the office. Some features required esoteric commands to function, and the system was painfully slow. Developers played Halo or watched Amazon Prime Video while waiting for Lumberyard to process art or compile code, several former employees say. A common refrain around the office, according to a former employee: “Lumberyard is killing this company.”
From the outside, though, Amazon’s game operation was scaring competitors. It had some of the world’s brightest minds, a deep budget and the most sophisticated internet infrastructure on earth. It was building a proprietary engine, and at TwitchCon in 2016, Amazon said it was working on three new games: Breakaway, Crucible and New World. (Nova, the League of Legends copycat, was still under lock and key.) Over the coming years, all but one would be canceled.
In the summer of 2018, Frazzini nabbed his biggest hire yet. Christoph Hartmann, the new vice president of game studios at Amazon, had spent two decades at Take-Two Interactive Software Inc., where he published blockbusters such as BioShock and Mafia. It was a somewhat controversial hire, as Hartmann was also responsible for some high-profile failures, such as The Bureau: XCOM Declassified, which led to the developer’s demise.
Still, Frazzini positioned Hartmann as a fixer to staff. Hartmann loosened the mandate to make everything with Lumberyard, and some teams began prototyping games using Unreal Engine. Hartmann also pressed Amazon to publish titles made by other companies. It signed one such deal with the Korean publisher Smilegate to release a game in 2021.
Soon after Hartmann’s appointment, Amazon came close to capturing a much brighter star in the media world. The company informed staff in 2018 that Jason Kilar, a former Amazonian and onetime Hulu CEO, was rejoining to oversee the Amazon games division and presumably become Frazzini’s boss, say two people who were there at the time. For unexplained reasons, Kilar never arrived. (The almost-hiring was first reported by technology news site the Information.) In April 2020, he joined WarnerMedia as CEO.
With Frazzini still at the helm, the game studios began aligning themselves with Amazon’s far more successful film and television properties. The company said in 2019 that it was working with the Chinese developer Leyou Technologies Holdings Ltd. on an online game based on the Lord of the Rings as a complement to the upcoming Prime Video series. That same year, Amazon released its first-ever console game, a racing simulation based on the Prime Video show The Grand Tour, a car review series starring the original British stars of Top Gear. The game bombed. It drew so few players that Amazon took the unusual step of removing it from storefronts a year later.
The embarrassments kept coming. In May 2020, Amazon released Crucible, the hero shooter inspired by Overwatch. “One of the things that we hear most often from people who try Crucible is that it feels unique,” Frazzini said in an interview at the time.
Gamers weren’t interested. Reviewers at IGN called it “tedious,” and PC Gamer declared: “Amazon’s long-awaited hero shooter wasn’t worth the wait.” Twitch didn’t offer much support, either. A week after release, fewer than 1,000 people were watching Crucible videos on Twitch. Former employees of the live-streaming unit say Twitch was often reluctant to alter a game’s fate through promotion. After all, Twitch can’t save a game nobody wants to play. Amazon pulled Crucible from wide release in June and then killed it in October.
Amazon shifted many of the Crucible developers to work on New World. The project was originally pitched as a survival game in which people would play as colonists in a fictional version of 1600s America, fighting enemies that looked a lot like indigenous people. The original code name for the game was Roanoke, named after Sir Walter Raleigh’s failed settlements in the 16th century. When developers at Amazon pointed out to Frazzini’s deputy, Patrick Gilmore, that the setting and villains could be considered racist, he expressed disbelief, according to two people who worked there. Gilmore didn’t respond to a request for comment. The disagreement, former employees say, fit a pattern of executives neglecting advice from staff.
Developers eventually removed the Native American imagery from New World. The game was set for release this past August, but after Crucible’s scathing reception, Amazon pushed the debut back to this year. Still, several Amazon employees are optimistic about the title. It has received positive buzz from streamers and is the kind of ambitious project Bezos is known to support, employees say.
Frazzini is fond of saying anything can be measured, according to two people who worked closely with him. This philosophy, common in Silicon Valley, is anathema to game industry veterans. “Nobody who’s successful starts with metrics,” says Seamus Blackley, who helped design the original Xbox. The Microsoft game console was the last winning incursion into the video game market in the past two decades, despite expensive attempts including Apple Arcade, Facebook Oculus and Google Stadia.
Fairly or not, Frazzini gets a lot of the blame for Amazon’s gaming failures at a time when overall spending on video games during the coronavirus pandemic is up significantly. After the cancellation of Crucible, an outgoing employee entered the Seattle office to collect personal belongings and found a crass message in big blue letters scrawled on a whiteboard: “Fraz is cancer.” Orbiting the note was a handful of “+1”s written in assorted colors.
The biggest new game product from Amazon has nothing to do with Frazzini. Amazon Luna lets subscribers instantly play console-grade games without the need to buy pricey hardware or wait hours to download a large file. The team released a version for Android devices in December, and the service remains in early access limited to customers who request an invitation.
The Luna project is overseen by David Limp, who runs the devices division that produces the Echo and Kindle. In its current iteration, Luna offers more than two-dozen games, none of which are made by Amazon Game Studios. “I hope that they have a hit,” Limp said in an interview in September. “But in addition, I think it’s equally important for us to build a system, and I think Luna has the starting points for that.”
In many ways, the approach to games mirrors the one that eventually led Amazon to some success in Hollywood. It tried a bunch of different things—develop a streaming service, set up a studio, produce TVs and films, build a set-top box—and selected an Amazon insider, Roy Price, to run it. Price produced a handful of bad shows before Transparent won Amazon its first Golden Globe in 2015. Then came Oscars for Manchester by the Sea and more Golden Globes for The Marvelous Mrs. Maisel. Price was ousted in 2017 over sexual harassment allegations, and Bezos took the unlikely step of looking outside of the company, to Hollywood, for his replacement, Jennifer Salke. She cemented Amazon Prime Video as an important part of the company’s business strategy. People sign up for Prime to watch The Boys, and they buy more stuff on Amazon.
Amazon could still do the same in gaming. Luna demonstrates a continued commitment, as do the investments in New World and as-yet-unannounced projects including a secretive new game from Smedley, who laid the foundation for massively multiplayer online games in the late 1990s with EverQuest. After Amazon’s misadventures in gaming, there’s at least one passage from the leadership principles the company will still hope to prove: “Leaders are right a lot.”
Written by Jason Schreier.